Structure and predictive functional profiling of microbial communities in two biotrickling filters treated with...

Two biotrickling filters were operated in continuous (BTF1) and discontinuous (BTF2) modes at a constant empty bed residence time of 60 s for 60 days. From day 60, the operation mode of each BTF was oppositely switched. Higher removal efficiencies of five aromatic pollutants were recorded with BTF1 (> 77.2%). The switch in the operation mode did not alter the removal performance of BTF1. Comparatively, BTF2 was not successfully acclimated in the discontinuous operation mode. Once the mode had been switched to continuous mode, the removal efficiencies of BTF2 on all pollutants increased drastically and finally exceeded the values observed in BTF1, with the single exception of p-xylene. Principle coordinate analysis and analysis of similarities (ANOSIM) showed that the structure of the microbial communities differed considerably between both BTFs (R = 1.000, p < 0.01) as well as before and after the switch in BTF2 (R = 0.996, p < 0.01). The random forest model demonstrated that Mycobacterium, Burkholderia, and Comamonas were the three most important bacterial genera contributing to the differences in microbial communities between the two BTFs. Metagenomics inferred by PICUSt (phylogenetic investigation of communities by reconstruction of unobserved states) indicated that BTF2 had high degradation potential for aromatic pollutants, although those genes involved in biofilm formation were less active in BTF2 than those in BTF1.